Dissolution of Potash Ore in Solutions of High Sodium Chloride Concentration RG. Gilliesl, D.N. Madge* and CA. Shook* Depatfmenf of Chemical Engineering, University of Saskatchewan, Saskatoon S7N OW, CANADA An experimental investigation of the dissolution of KCI-containing ore in brines of high NaCl content has been conducted. The dissolving su@ace was vertical and motion of the solution was due only tofree convection. The effects of ore composition and brine NaCl concentrationwere studied for unsaturated brines using both reconstituted ore and raw ore slabs (11J cm high). These dissolution rates were consistent with previous free convection heat and maps transfer results, when the interfacial solution was assumed to be saturated with respect to both KCl and NaCl (the invariantpoint composition). Dissolution by brines saturated with respect to NaCl was investigated with respect to the effects of crystal size in the slabs, solution KCl concentration, temperature and time. Dissolution rates were much lower in these cases and appeared to be controlled by a surface reaction. The mechanism of dissolution appears to involve penetration of the slab bv the dissolvinz solution. Introduction Potash has been mined in Saskatchewan since 1958 and at present it is the most important mineral produced in the province. Since vast quantities of NaCl-rich/KCllean railings have been left on the surface, the possible dissolution of these deposits is a matter of concern. With growing interest in solution mining (one of the original shaft mines has been converted to a solution mine) and with recent concern about migration of aqueous solutions underground, the dynamic behaviour of the system NaCl-KCl-%O is of interest. The present investigation was undertaken to examine ore dissolution rates in aqueous solutions containing KCl and NaCl. The region of greatest interest was that of relatively high NaCl concentrations. As in a previous investigation (Husband and Oszahin, 1967) the variability and insoluble content of the natural ore required reconstituted ores to be fabricated in an attempt to obtain reproducible experimental conditions. Dissolution rates were to be determined using equipment of laboratory scale. In view of the complexity introduced by the presence of two species dissolving simultaneously,the experimentalconditions were selected to provide a regime whose hydrodynamic conditions are relatively well understood. The particular situation was free convective dissolution of vertical slabs immersed in a large quantity of the solution. ~ Saskatchewan Research Council, Sashtoon, CANADA Kilborn Western Inc., Sashtoon, CANADA * Authorfor correspondence. 218 Dissolution of Potash Ore in Solutions ofHigh Sodium chloride Concentratwn Since brines saturated with respect to NaCl can be produced in a wide variety of physical situations which occur above and below ground, the experimental program was also designed to examine the behaviour of ores in contact with a brine of this type. Both natural and reconstituted ore slabs were used in this phase of the investigation. (a) Free Convective Dissolution of Vertical Surfaces Molecular diffusion of a single solute (component 1) is described by Fick's Law: where u is the component of the mean velocity of the solution in the direction of N. For transfer in the y direction: Component 2 is the other constituent of the binary mixture. For liquids, because the mass density is nearly constant, an appropriate mass transfer rate equation for engineering design is: where k, is the mass transfer coefficient which depends, among other factors, upon the diffusivity and the hydrodynamic conditions in the vicinity of the dissolving surface. At low rates of mass transfer, single component mass transfer and heat transfer are strongly analogous; the relationship is implicit in the conservation laws for mass, momentum and heat, and in the physical laws governing the diffusion of heat and mass by molecular motion. In laminar free convective heat m s f e r from a vertical plate, it can be shown that: Nu = f (Gr, Pr) (4) where the Nusselt number, Nu = h uk;L is the length of the surface; k is the thermal conductivity; Pr is the fluid Prandtl number, FV = pC@; C, is the heat capacity. The Grashof number (Gr) is the ratio of buoyancy forces to inertial forces: The mass transfer relationship which is analogous to Equation (4) is: Sh" = f (Gr, Sc) where the Schmidt number, Sc = NpD,; and the Sherwood number, Sh" = Isp L/D1. The product (Gr Sc) is the Rayleigh number for mass transfer. The coefficientlip is the limiting value of the mass transfer coefficient at very low rates. For highly 219 R.G. Gillies, DN.Madge and CA. Shook soluble substances, the relationship between these mass transfer coefficients is given by: Although the difference between k and kqo is recognized in forced convection mass transfer, earlier free convective dissolution studies often ignored the difference. Experimental data have been presented in the form of Sherwood numbers defined as: and plotted in terms of the Rayleigh number (Gr Sc). These resdts are usually compared to those obtained in heat transfer, for which an accepted correlation for laminar boundary layers is: Nu = 0.667 (Gr Pr)o-25 (6) Equation (6) can be derived by considering the velocity and thermal boundary layers in free convection. If the boundary layer becomes turbulent, then the correlation of Churchdl and Chu (1975) is appropriate: Nu0q5 = 0.825 + 0.387 (Gr Pr)'I6/[1 + (0.492/Pr)9/16]8''27 (7) 100000 10000 -- - Churchill h Chu Husband Experiment - 1000 100 I~ I I1111 10 1 ; II 11 Ills Id m GrSc 1 1 1 T IS 111- 18 10 Figure 1 . Comparison of the experimental Shenvood numbers of Husband and Shook (invert sugar dissolution) with the correlation of Churchill and Chu. If the mass transfer process is one of diffusion controlled by free convection at the surface, then Equation (7) can be used to predict mass transfer coefficients with Nu = Sh and Pr = Sc, in both laminar and turbulent regions. The applicability of the Churchill and Chu (1975) correlation to free convection mass transfer is shown in Figure 1, which compares the experimental results of Husband and Shook (1970) 220 Dissolauion of Potash Ore in SolutiorrS @High Sodium Chloride Concenlratwn with the correlation. Husband's experiments used invert sugar slabs for which Sc = 2.89 x 106 and xls = 0.23, and agreement with the correlation is seen to be excellent. In these experiments (Husband, 1969), slabs with L values ranging between 7.6 cm and 6.1 m were used and it was noted that the precision of the measurements improved as L increased. The transition from a laminar to a arrbulent boundary layer occurs at a value of (Gr Sc) near lo9 for low Pr fluids, while Husband (1969) found that it occurred at (Gr Sc) near 1013with the invert sugar slabs. If a chemical reaction occurs at the surface as part of the dissolution process, mass transfer coefficients predicted by Equation (7) will be higher than the experimental values. Evidence for a surface reaction in forced convective dissolution of pure KCl into NaCl solutions and pure NaCl into KC1 solutions was given by Simon (1981). (b) Potash Ore Dissolution by Unsaturated Brines When two salts are present, Equation (1) can be applied to each component at any point on the dissolving surface. However, Equation (2) requires modification (eg. Fujita and Gosting, 1960): J, = -D, acl/ay - ~ 1acday 3 and where components 1 and 3 are identifed as NaCl and KCl respectively. The crossterm diffusion coefficients D13 and D31 are approximately 15 to 20%of D, and D3. If Equation (3) is to be used to calculate dissolution rates, the surface compositions (xis, x3J are required and these will vary from point to point. Figure 2 shows the loci of isothermal saturation compositions in the form of mass ratios of NaCkwater and KC1:water in the region of high NaCl concentrations. Curve AB shows solutions saturated with respect to NaCI; the negative slope indicates the common-ion (Cl-) effect. Curve BC applies to solutions saturated with respect to KCl. At the junction (B), the invariant point, the solution is saturatedwith respect to both salts. In the coordinate system of Figure 2, a dry mixture of NaCl and KCl would be represented by a point at infinity, lying on a line passing through the origin, with a slope determined by the ratio of NaCl to KCl. In interpreting their dissolution rate measurements, Husband and Oszahin (1967) appear to have defined the surface compositions (xh and x 3 by the intersection of this line with curves AB and BC. This is certainly a plausible engineering simplifying assumption but it is not the only one. For example, when the total perimeter of the junctions between NaCl and KCl crystals per unit area of dissolving &ace is high, it would be equally appropriate to assume that the mean d a c e concenmtionfor the slab as a whole was that of point B. 221 R.G. Gillies,DN.Madge and C A . Shook 0.4 I * Q) CP Y 0.3 1 0 u 0.2 7 CP Y 0.1 I 0.0 1 0.j I 0.2 I 0.3 kg KCl/kg water Figure 2 . Phase diagram for the system NaCl-KCl-water at 29OC. (The solid points depict the solutions used in the short-term experiments). Husband and Oszahin (1967) reported good agreement between their three measured initial dissolution rates, as determined with reconstituted ore slabs containing approximately 50% KCl, and the predictions made with Equation (6). In their experiments, L was approximately 21 cm. In order to obtain this good agreement, it was necessary to extrapolate measured rates to overcome the effects of turbulence and roughness which developed on the surface. (c) Dissolution by Saturated Brines According to the mechanism implicit in the method of Husband and Oszahin (1967). brine in contact with ore will eventually become saturated with respect to one of the salts. Once saturation occurs, according to this view, dissolution should cease. This situation was examined experimentally by Taylor et al. (1967). Using ore samples exposed to brine at their lower surfaces, deposition ('blinding') of NaCl on the surface during KCl dissolution was observed when the solution was sahrrslted with respect to NaC1. Simon (1981) found that deposition of KC1 on an NaCl crystal occurred very rapidly, when the solution KC1 concentration was high, even when the solution was undersaturated in both salts. Similarly, NaCl was deposited very rapidly on a KCl crystal from a solution of high but undersaturated NaCl concentration. These observations suggest that dissolution in brines of high salinity could decrease rapidly with time, or cease altogether. When an ore is placed in contact with a saturated brine, the density difference in the Grashof number is very small. In this situation the mass transfer mechanism could be significantly different from that which applies to unsaturated brines. A separate set of experiments was planned to examine the behaviour of ore in contact with a saturated brine. In these experiments, weight losses as a function of time were to be monitored to examine whether or not dissolution occurred, and the relative importanceof the 'blinding' phenomenon. 222 Dissolution of Potash Ore in Solutwns of High Sodium Chloride Concentration Experimental Procedures (i) Density and Viscosity measurements Densities and viscosities as functions of solution compositions were determined experimentally for the KC1-NaC1-water system. The experiments were performed at 25°C and 29°C. The temperature of the entire test chamber was held constant (at either 25°C or 29°C) to eliminate temperature differences. Brine samples were prepared by dissolving accurately weighed amounts of reagent-grade KCl and NaCl in distilled water. Prior to testing, the brine samples were allowed to stand in sealed containers for at least 16 hours after preparation, to ensure that the temperature had equilibrated. The densities were measured by determining the weight and volume of the brine samples using 100 ml volumetric flasks. Capillary tube viscometers were used to measure the viscosities of the brine samples. Blank determinations were performed using distilled water to verify the flask volumes and the calibrationsof the capillary tube viscometers. (i) Ore Sample Preparation The dissolution-rate tests were performed on relatively small laboratory specimens of potash ore. Most of the tests were conducted on reconstituted ore samples prepared by combining measured mass fractions of KCl crystals and NaCl crystals. The crystals were obtained from the dense media flotation circuit at the International Minerals and Chemical Corporation plant in Esterhazy, Saskatchewan, and their purities ranged between 95% and 98%. They were screened to provide come (A mesh, ds0 = 6 mm) and medium sized (-4+8 mesh, ds, = 3 mm) fractions. The fine crystal fraction consisted entirely of -8 mesh material. The medium-sized crystals were used for the majority of the tests. The coarse and fine size fractions were used for one series of tests designed to examine the effects of crystal size. The crystal mixtures were pressed using a stainless steel mould and a 50 ton hydraulic press. The reconstituted ore specimens were formed by applying a pressure of 33 Mpa over a 24 hour period. For most tests, the pressed ore samples were cut and trimmed to nominal dimensions of 12 cm x 10 cm x 2 cm thick. One 12 cm x 10 cm face was available for dissolution and all the other surfaces were sealed with a moisture resistant epoxy sealant. The exposed face was sanded to a smooth finish. Some samples of raw potash ore (48% KCI, 49% NaC1) were also used during the experiments. These ore samples were cut from 200 mm diameter core samples provided by Cominco Ltd. The raw ore samples were prepared in the same manner as the artificial samples so that all surfaces except one were sealed, and the exposed surface was sanded to a smooth finish. (ui) Data Acquisition System Brine temperatures were measured using type-T thermocouples. Immersed specimen weights were measured using 0 to 500 g electronic load cells supplied by Cell Builders Inc. The load cells were temperature compensated over the range 0 to 65°C and their precision estimated as 0.1 g in isothermal conditions. A Seimecric Model M8082A analog-to-digital converter and a Compaq XT personal computer were used to monitor and record the instrument readings. Most of the tests were conducted at 29"C, with a final set of measurementsat 25,30 and 35°C. 223 R.G. Gillies,D N . Madge and C A . Shook (iv) Short-Duration Tests Tests were performed to determine weight-loss rates for reconstituted potash ore specimens formed using medium sized crystaIs. AU the tests were conducted at 29°C. Four grades of ore (0,40,60 and 100%KCl) were tested. Samples of raw ore were also tested. The samples were suspended in 4 litre brine samples which contained 7% KCl (by mass) and various levels of NaCl. The NaCl concentrations were 15, 18.21 and 22.7% by mass. The latter brine sample was saturated with respect to NaCl. The actual weight-loss rates were determined using the measured immersed weight of the specimen, the brine density (p) and the solids (ore)density @s> using the relationship: Actual Weight = Immersed Weight / [l - @/ps)] Most of the weight losses were measured for approximately 30 minutes. For the tests with the saturated brines, &he measurements were continued for a longer period of time (nominally 20 hours). A layer of mineral oil was placed over each brine sample to prevent evaporation during the 20-hour long tests. During the test, some dispersed solids collected at the bottom of the beakers. Some of this sediment was insoluble material released from the ore block during dissolution, and some consisted of salt crystals which had become detached from the ore block during the dissolution process. The solids were collected at the end of each test by decanting and filtering the brine samples. The solids were submitted to the Pilot Plant Laboratory of the Potash Corporation of Saskatchewanfor analysis. The tests were initiated by lowering the ore slab gently into the beaker containing the brine solution. Any air trapped within the slab escaped in the first two or three minutes. This effect, and the fact that the slabs could not be lowered in a completely reproducible manner, required the data for the first few minutes to be rejected. Thereafter, the weight-loss rates in the short-term tests were stable with respect to time. The test conditions and weight loss results are summarized in Table 1. Of particular interest is the fact that eddies were not observed near the surface of the slabs when the most concentrated solutions were used. The solution compositions are shown as points D, E, F and G in Figure 2. (v) Long-Duration Tests with Saturated Brine Longduration tests were performed to determine weight-loss rates for potash ore specimens formed using coarse, medium and fine crystals. In this case, the composition was 40% KCl and 60% NaCl. A raw ore sample was also used. The specimens were nominally 12 cm x 10 cm x 5 cm thick. The back and the two sides of each specimen were sealed with moisture resistant epoxy so that the top, bottom and one vertical face were available for dissolution. The tests were canied out using brine which contained 7% KCl and 22.7% NaCl (saturated with respect to NaC1). Each specimen was immersed in a relatively large tank of brine (48 lilres) to minimize the change in brine concentration during the test The test temperature of 29°C was maintained by regulating the temperature of the 224 Dissolutionof Potash Ore in Solutwm of High Sodium Chloride Concematwn Table I . Summary of test conditionsfor short-duration experiments (temperature = 29°C; medium sized crystals). lo0 7.0 15.0 0.5 150 100 7.0 18.0 0.5 70 A B loo 7.0 21.0 0.5 29 C 100 7.0 22.7 20.5 3.1 C 60 7.0 15.0 0.5 270 &B 60 7.0 18.0 0.5 98 A. B 60 7.0 21.0 0.5 '34 C 60 7.0 22.7 205 1.6 C 40 7.0 15.0 0.5 250 AB 40 7.0 18.0 0.5 120 AB 40 7.0 21.0 0.5 14 C 40 7.0 22.7 20.5 0.7 C 0 7.0 15.0 0.5 170 A. B 0 7.0 18.0 0.5 34 A, B 0 7.0 21.0 0.5 9 C 0 7.0 22.7 20.5 0 E Raw on 7.0 15.0 0.5 I10 A B. D Raw on 7.0 15.0 0.5 170 A B. D Raw olt 7.0 18.0 0.5 81 A, B. D Raw OIC 7.0 21.o 0.5 26 C, D Raw ore 7.0 22.7 0.5 0 E A. B Key for Observations: A. B. C. D. E. Rapid dissolution o c c u d and density cumnt eddies were visible to the naked eye. Significant roughening of the surface occurred during thc test A small amount of surface rougbcning occurrrd during thc test A film of insoluble mated colleaed at the dissolving face. The om specimen did not change noticeably during the test. room. Layers of mineral oil were placed above the brine samples to prevent evaporation from the tanks. Dispersed solids were collected from the bottom of the tanks and submitted for analysis. The test conditions and dispersed solids (sediment) analysis are summarized in Table 2. 225 R.G. Gillies.DN.Madge and CA. Shook Table 2. Summary of test conditions for long-duration experiments designed to examine the effects of ore crystal size (T = 29°C). On Quality crystal Jize Lengthof Observations BrincComposition(mus) % KCI % NaCl (See key following Table 1) Test-(hrs) % KCI coarse 40 732 7.0 22.7 B Medium 40 1176 7.0 22.7 B Fine 40 732 7.0 22.7 B -48 I224 7.0 22.7 B, D ROW on? Table 3. Swnmary of test conditwm for long-duration tests to examine the effects of brine quality and temperature (ore quality = 40% KCl ;medium sized crystals). 2s 8.0 22.2 237 1.8 4.9 15.8 19.3 B 25 9.0 21.6 237 I .? 4.6 82.1 13.3 B 25 10.0 21.1 237 2.2 17.8 10.6 11.6 C 30 8.0 22.2 311 I .4 2.3 915 6.5 B 30 9.0 21.7 311 1.4 3.2 863 11.3 30 10.0 21.2 311 2.4 s.1 43.4 22.4 B B 35 8.0 223 160 1 .S 6.6 88.1 5.5 B 35 9.0 21.1 160 3.1 2.2 91.1 4.6 B 35 10.0 21.2 160 3.4 113 l3.6 12.6 C (vi) Long-Duration Tests for Temperature and Brine Composition Effects Three series of long duration tests were performed to examine the effects of brine composition and temperature on the potash dissolution process for saturated brines. The tests were conducted at temperatures of 25.30 and 35OC using brines with KCl concentrations of 8,9 and 10%by mass. In each case, the brines were saturated with respect to NaC1. The brine samples were contained in 4000 ml beakers which were placed in a large temperature controlled water bath. A layer of mineral oil was placed over each brine sample to eliminate evaporation. The ore specimens contained 40% KC1 and 60% NaCl and were prepared by pressing medium sized crystals. The specimens were trimmed to nominal dimensions of 12 cm x 10 cm x 2 cm thick. One 12 cm and 10 cm face was exposed to the brine and all other faces were sealed with epoxy. The exposed face was sanded to a smooth finish. 226 Dissolutim of Potash Ore in Solutions $High Sodium Chloride Concentration The tests were conducted for several days to establish long term trends. The test conditions and analysis of dispersed solids material are summarized in Table 3. Experimental Results and Discussion (a) Viscosities and Densities of Solutions of KCl and NaCl The relative Viscosities (dynamic viscosity divided by the viscosity of water at the same temperature) became more sensitive to KCl content as the NaCl concenmtion of the solution increased. Densities of the solutions were primarily functions of the total solute molarity. Data for these solutions can be obtained from the authors. (b) Unsaturated-Brine Dissolution Rates The mass-transfer rate equation (Equation 3) is known to be appropriate when a single component (pure KCl or NaCl) dissolves. In Figure (2), for a solution with bulk composition F (x,J. the surface concentration (xb) is given by the interseCtion of line FH with AB, or FJ with BC. In the absence. of surface mction effects, the mass transfer coefficients should be consistent with those obtained in previous friee convection experiments. If Equation (3) is to be used to calculate mass transfer rates when two components dissolve, then the mass transfer coefficient (kx) must be defmed. In principle, k, should be different for the two species because, on an area-averaged basis, there are two concentration boundary layers (one for each solute), although there is only one velocity boundary layer. However, the diffusivities of NaCl and KC1 differ by less than 25% so that in this particular case a simpler approach could be used. In converting experimental dissolution rates to mass transfer coefficients and Sherwood numbers, Equation (3) is applied as if the ore were a pseudohomogeneous substance. Specifically, when two solutes are present: a) Nl is the total dissolution flux (lcg/m2s); b) xl= is the sum of the mass fractions of the two salts in the bulk of the solution; c) xls is the total solute mass fraction at the surface, computed for point B. This is an alternative to the method of Oszahin and Husband (1967). Using the ores containing 40% and 60% KCl, and a solution with composition F, the method of Husband and Oszahin predicts the surface compositions determined by the intersection of line FK with AB, or F'L, with BC. When Sherwood numbers (Sh) or Schmidt numbers (Sc) were calculated, mean diffusivities for the solution were computed from the diffusivities of pure NaCl and pure KCl in water assuming a linear mixing rule. Values of these diffusivities at 25°C are given by Harned and Owen (1958) and were corrected for the effect of temperature using the activation energies quoted by Simon (1981). Viscosities and densities were interpolated from the experimental values. Unless noted otherwise, diffusivities, Viscosities and densities were determined from the arithmetic mean of the surface and the bulk concentration. These definitionsobviously involve additional approximationsin two-component dissolution. Comparison of the two-component results with single-component dissolution should help to assess these simplifyingassumptions and definitions. 227 R.G. Gillies, D.N. Madge and CA. Shook (c) Single-ComponentDissolution by Unsaturated Solutions The experimental N, values were used to obtain k, values, which are shown in Figure 3 as values of the Sherwood number plotted against the Raylei& number. At the higher solution concentrations (points E and F in Figure 2), where the lower Rayleigh numbers occur, the experimental values are close to those predicted by the laminar boundary layer equation. However, for the brine with the lowest concentration (point D in Figure 2), the S h e r w d numbers are significantly higher, so that the measurements are closer to values predicted by the correlation of Churchill and Chu. Considering the observations shown in Table 1, it seems likely that the transition from laminar to turbulent flow plays an important role for specimens of this length. This probably explains the relatively poor reproducibility (30% deviations) between replicate tests despite the precision of the weight loss measurements. This relatively poor reproducibility for small specimens was noted previously by Husband (1969). (d) Two-Component Dissolution by Unsaturated Brines These rates were expressed as k, values using Equation (5) and the assumptions stated previously. These results are also shown as values of the Sherwood number plotted against the Rayleigh number in Figure 3. The invariant point (point B in Figure 2) was used to calculate the equilibrium concentration (xis) in this case. The agreement with the singlecomponentSherwood numbers is satisfactory. The points shown at Rayleigh numbers near l O I 3 are the dissolution rate measurements of Husband and Oszahin (1967) interpreted in the same way as the mixture data obtained in this investigation. In calculating the Sherwood and Schmidt numbers for the measurements of Husband and Oszahin, the effects of temperature on the physical properties were also calculated using the activation energies quoted by Simon (1981). Husband and Oszahin's inference that their dissolution process was laminar remains valid when their data are interpreted using the invariant point as the surface condition. This similarity in the values of the mass transfer coefficients calculated from two different definitions of the interface condition occurs because they used water as the dissolving fluid. If the method of Husband and Oszahin were used to determine the concentration xls forthe brines used in the present investigation, the mass transfer coefficients and Sherwood numbers for the mixtures would be significantly higher than those which apply for the pure salts. These higher coefficients occur because the concentration difference between the solution and point B in Figure 2, is greater than that between the point and the saturation lines AB and BC along lines such as FK or FL. It therefore appeats that the invariant point composition is useful for estimating xIsr when the procedure described previously is employed. Although the range of Rayleigh number is rather restricted, the evidence suggests that the laminar boundary layer and Churchill-Chu correlation provide useful indications of the magnitudes of free convective mass transfer coefficients for ores containing 40 to 60% KCI. The experimental Sherwood numbers are actually somewhat higher than the limits implied by these correlations. These deviations could be due to effects which are neglected in a pseudo-single component approach and which are shown in Equations (5)and (8). 228 Dissolution of Potash Ore in Solutwns @High Sodium Chloride Concematwn 10000 5 v, 1000 1 Figure 3. Experimental Sherwood numbers as functions of Rayleigh number for dissolution of pure KCl, pure NaCl and mixtures of KCl and NaCl in brines D , E and F (see Figure 2). (Results of Husband and Oszahin are also shown), 10000 Raw Ore 100 : I 1111111 1 I11111 I I l m l I I Ild 229 R.G. Gillies,DN.Madge and CA. Shook The experimental measurements of Husband and Oszahin (1967) are of considerable importance in establishing the mechanism of the dissolution. Their temperature range (21.1OC to 54.4OC) was relatively wide, and the modest change in rate which they observed shows that a surface reaction of the type detected by Simon (1981) does not play an important role in two species free convection dissolution. (e) Dissolution by Saturated Brines The short-durationtest results in Table 1 show that rates of dissolution by saturated NaCl solutions were very low. Even when the test period duration was increased to 20.5 hours, the changes in weight were probably too small for rates to be measured reliably (the pure NaCl slab slightly increased in weight). These unsatisfactory results led to the longduration tests which are summarizedin Table 2. The first set of longduration tests examined the effect of sample heterogeneity by using reconstituted ores with different crystal sizes and a sample of raw ore. Although there was some difference between the results, the rates were similar and show that blinding did not cause dissolution to cease. The insolubles in the raw ore, which led to obvious surface deposits, only seemed to reduce the dissolution rate by about 50%. The rates gradually decreased with time. Because three surfaces were exposed to the brine in these tests, it was difficult to draw firm conclusions about the cause of the gradual decrease in the rate with time. However, it seems probable that it was due to the KCl content of the slab becoming exhausted. A further series of tests un&r conaolled hydrodynamic conditions was undertaken, with the contact time somewhat reduced and the KCl concentration increased. In the second set of saturated-brine dissolution tests shown in Table 3, the specimens were vertical so that the orientation was identical to that used in the unsaturated-brine dissolution measurements. In these experiments, there was an initial period of 25 to 100 hours in which the dissolution rates were relatively high, i.e. consistent with free convective dissolution under conditions of low Grashof number. Thereafter, dissolution appeared to continue at a very low but apparently constant rate. The variable initial rates are understandable considering that the brine compositions were prepared at saturation compositions determined from interpolation of the phase diagram (Linke, 1965), and that the slabs released small quantities of air when fmt immersed in the solutions. In view of their possible relevance to solution mining, the low ultimate rates were estimated by ignoring the initial periods. To interpret these rates, the mass transfer pracess in a brine saturated with respect to NaCl was assumed to be: NaCl(,) + KCl(,) = NaC4,) + KC+,,) (9) Using rate equations to describe transfer between the saturated surface (s) and the bulk (m) of the solution, and omitting the subscript x from the mass transfer coefficients: 230 Dissolution of Potash Ore in Solutwns @High Sodium chloride Concematwn the net rate of mass decrease is given by (NKAK - NNaANa), where AK and ANa denote the areas at which the transfer takes place. Thus: where The term in the square brackets in Equation (11) is the difference between the mass fractions at the invariant point (s) and that in the bulk of the saturated solution & is the effective transfer coefficient. Values of k,were calculated from the experimental weight-loss rates (lcg/m2 s). They are very low compared to the free convective mass transfer coefficients in unsaturated brine, presumably because dissolution actually takes place at a very small fraction of the surface area A. However, like the rates themselves, these saturated-brine mass transfer coefficients are strongly temperature dependent. Figure 5 is an hhenius plot of the k,values; the activation energy is of the order of 14 kcallmole, which is much higher than that associated with a diffusion mechanism. The activation energies observed by Simon (1981) for pure KCl and pure NaCl dissolution surface reactions were lower than that implied by Equation (9). It should be noted that Simon used an equation similar to Equation (10) to define the rate constant for the reaction process. (-); 1/T ( K-’) Figure 5. Dissolution rate constants (k,,,)for NaCl-saturated brines versus 1IT. To interpret the low rates of mass transfer in the presence of a saturated NaCl solution, we must reconsider the weight loss results. A slab of initial mass approximately 1.2 kg, containing approximately 480 g KC1, lost approximately200g as the KCl was removed and NaCl was deposited. During this time the surface roughened and the mass transfer process was not dominated by free convection in 231 R.G. Gillies,DN.Madge and CA. S h k the solution. Therefore, it Seems likely that the dissolution ~ ~ O C proceeds ~ S S within the slab rather than at the surface. According to this view, the solution must penetrate the slab, presumably at crystal junctions, dissolve KC1 and deposit NaCl in accordance with Equation (9). A solution having composition near that of the invariant point must return to mix with the bulk of the solution. Because of differences in the rates of dissolution and the specific volumes of the two solid species, the porosity of the slab increases with time. This porosity would appear as an increase in roughness. Hence, the low k,,, values are due in part to the relatively small areas at which simultaneous dissolution and deposition take place within the slab (per unit area of slab surface). However, the process must be complex because increasing the contact area between KC1 and NaCl (by decreasing the crystal sizes) did not increase the overall rate. Conclusions and Recommendations The experimentssuggest that: 1) Free convective dissolution of ores in unsaturated brines occurs at rates comparable to those predicted by accepted correlations, if the interfacial compositionis assumed to be that of the invariant point. 2) Ore containing 40% KCl dissolves in saturated brine at rates which are relatively low, after an initial phase lasting up to 100 hours. These rates do not decline significantly with time. The dissolution process in this case appears to be controlled by a reaction, rather than diffusion. 3) Although the insolubles cause dissolution rates to be somewhat lower for raw ore than for reconstituted ore, the effect is only moderate. Since the cost of preparing reconstituted ore slabs is relatively high, further experiments with reconstituted ore are probably not justified. The physical process by which dissolution into a saturated solution occurs seems to be worth investigation. It is our view that the saturated solution must penetrate the slab and invariant concentrationbrine must leave it simultaneously. Although use of the invariant point composition seems to allow mass transfer rates to be estimated for ores containing between 40 and 60% KCI, it is not clear if this assumption would be justified for other compositions and it is obviously not appropriate for the pure salts. The behaviour of lean ores containing between 0 and 40% KCI, and rich ores containing between 60 and 100%KC1, should be examined with respect to Conclusion 1 above. Acknowledgments "his investigation formed part of a study conducted by Kilbom Western Inc. for the Saskatchewan Potash Producers Association. Mr. P. Schergevitch of the Saskatchewan Research Council conducted the experiments. The authors are indebted to the SPPA for permission to publish the results. Helpful discussions with the SPPA management committee, Messs. A. Banks and D. Thompson of Kilbom and Dr.W.H.W. Husband are acknowledged. 232 Dissolution of Potash Ore in Solutwm @High Sodium Chloride Concentration Nomenclature A k e a of lab surface CL2, C Concentration(ML") D Diffusivity (L2T1) Acceleration due to gravity (LT-9 g h Heat transfer coefficient m3) FIUX relative to mean velocity (ML-~T') J Effective mass transfer coefficient (LT') k, Is, Mass transfer coefficient (LT-') Vertical dimension of dissolving slab Q L m Mass of slab 0 N Mass flux relative to fvred coordinates (ML'2T-1) t Time 0 X Mass fraction solute Distance perpendicular to surface (L) Y P viscosity (ML-IT') P Density (ML-3) Subscripts av Mean value of solutions S Surface value (invariant composition assumed) X Implies concentrationexpressed as mass fraction 0 Bulk solution value References Churchill, S.W. and Chu. H.H.S. 1975. Correlating Equations for Lammar and Turbulent Free Convection from a Vertical plate. IN. J.Hear and Mass Tramfa, 18.1323-1329. Fujita, H. and Costing. L.J. 1960. A New Procedure for Calculating the Four Diffusion Coefficients of Three Component Systems from Gouy Diffusiometer Data. J . Phys. Chem., 64.1256-1263. Ham& H.S. and Owen,B.B. 1958. The Physical Chemistry of Electrolytic Solutions, 3rd Ed. American Chemical SOC.Monograph Series. Reinhold, New York. Husband, W.H.W. 1969. Free Convection Mass Transfer from Vertical Surfaces. P h 9 . Thesis in Chemical Engineering, University of Saskatchewan. Husband,W.H.W. and Oszahin, S. 1967. Rates of Dissolution of Potash Ore. Can. J . Chem. Eng.. 4,234-237. Husband, W.H.W. and Shook, C.A. 1970. An Experimental Study of Free Convection Mass Transfa h m High Vertical Surfaces to Liquids. Proc. 3rd Synpsium on Sail, Northem Ohio Geological Society, pp. 360-370. Linke, W.F. (Ed.). 1965. Solubilities of InorgMic and MetatOrganic Compounds, Vol. II, 4th Editwn. hexican Chemical Society, Washington, D.C. Simon,B. 1981. Dissolution Rates of NaCl and KCl in Aqueous Solution. J. C r y d Growrh, 12.789-794. Taylor. J.B.. Hunt, M.R., Despault, G.J. and Agyako, A.H. 1967. Selective Extraction of Potassium Chloride fiom Saskatchewan Sylvinite Ore. Can. J. Chem. Eng.. 4.105-108. Received 22 July 1993; Accepted after revision: 2 May 1994. 233

1/--страниц